Sign-tracking, a conditioned response in which animals engage with reward-predictive cues, is a powerful behavioral tool for assessing Pavlovian motivation. In rodents, it is most frequently studied via automatic readouts, such as deflections of levers that act as reward cues. These readouts have been immensely helpful, but they may not be ideal for some tasks and paradigms. For example, animals can show a range of sign-tracking responses to a lever cue that do not result in lever deflection, and a reduction in deflections when animals are exposed to an omission contingency (i.e., when lever deflection cancels reward) hides the fact that the animals are still sign-tracking in other ways. Here, we analyzed the behavior of sign-tracking animals through both video monitoring and automatic task readouts in Pavlovian conditioning. This analysis aided in the classification of sign-tracking animals and revealed that lever deflections do not result from any identifiable pattern of sign-tracking. We then used omission and extinction procedures to unmask detailed behavior changes that can only be detected with video data. Automated readouts showed similar reductions of lever deflection in both task conditions. However, detailed behavioral analysis revealed quite distinct behavioral adaptations to these conditions with sign-tracking decreasing entirely during extinction while many sign-tracking behaviors (biting, sniffing, etc.) seemed to remain persistent during omission despite the decrease in deflections. Detailed behavioral analysis was thus critical for capturing sign-tracking maintenance, persistence, and loss.